yt_dlp/extractor/awaan.py

   1 import base64
   2 import urllib.parse
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     format_field,
   7     int_or_none,
   8     parse_iso8601,
   9     smuggle_url,
  10     unsmuggle_url,
  11     urlencode_postdata,
  12 )
  13
  14
  15 class AWAANIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?show/(?P<show_id>\d+)/[^/]+(?:/(?P<id>\d+)/(?P<season_id>\d+))?'
  17
  18     def _real_extract(self, url):
  19         show_id, video_id, season_id = self._match_valid_url(url).groups()
  20         if video_id and int(video_id) > 0:
  21             return self.url_result(
  22                 f'http://awaan.ae/media/{video_id}', 'AWAANVideo')
  23         elif season_id and int(season_id) > 0:
  24             return self.url_result(smuggle_url(
  25                 f'http://awaan.ae/program/season/{season_id}',
  26                 {'show_id': show_id}), 'AWAANSeason')
  27         else:
  28             return self.url_result(
  29                 f'http://awaan.ae/program/{show_id}', 'AWAANSeason')
  30
  31
  32 class AWAANBaseIE(InfoExtractor):
  33     def _parse_video_data(self, video_data, video_id, is_live):
  34         title = video_data.get('title_en') or video_data['title_ar']
  35         img = video_data.get('img')
  36
  37         return {
  38             'id': video_id,
  39             'title': title,
  40             'description': video_data.get('description_en') or video_data.get('description_ar'),
  41             'thumbnail': format_field(img, None, 'http://admin.mangomolo.com/analytics/%s'),
  42             'duration': int_or_none(video_data.get('duration')),
  43             'timestamp': parse_iso8601(video_data.get('create_time'), ' '),
  44             'is_live': is_live,
  45             'uploader_id': video_data.get('user_id'),
  46         }
  47
  48
  49 class AWAANVideoIE(AWAANBaseIE):
  50     IE_NAME = 'awaan:video'
  51     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?(?:video(?:/[^/]+)?|media|catchup/[^/]+/[^/]+)/(?P<id>\d+)'
  52     _TESTS = [{
  53         'url': 'http://www.dcndigital.ae/#/video/%D8%B1%D8%AD%D9%84%D8%A9-%D8%A7%D9%84%D8%B9%D9%85%D8%B1-%D8%A7%D9%84%D8%AD%D9%84%D9%82%D8%A9-1/17375',
  54         'md5': '5f61c33bfc7794315c671a62d43116aa',
  55         'info_dict':
  56         {
  57             'id': '17375',
  58             'ext': 'mp4',
  59             'title': 'رحلة العمر : الحلقة 1',
  60             'description': 'md5:0156e935d870acb8ef0a66d24070c6d6',
  61             'duration': 2041,
  62             'timestamp': 1227504126,
  63             'upload_date': '20081124',
  64             'uploader_id': '71',
  65         },
  66     }, {
  67         'url': 'http://awaan.ae/video/26723981/%D8%AF%D8%A7%D8%B1-%D8%A7%D9%84%D8%B3%D9%84%D8%A7%D9%85:-%D8%AE%D9%8A%D8%B1-%D8%AF%D9%88%D8%B1-%D8%A7%D9%84%D8%A3%D9%86%D8%B5%D8%A7%D8%B1',
  68         'only_matching': True,
  69     }]
  70
  71     def _real_extract(self, url):
  72         video_id = self._match_id(url)
  73
  74         video_data = self._download_json(
  75             f'http://admin.mangomolo.com/analytics/index.php/plus/video?id={video_id}',
  76             video_id, headers={'Origin': 'http://awaan.ae'})
  77         info = self._parse_video_data(video_data, video_id, False)
  78
  79         embed_url = 'http://admin.mangomolo.com/analytics/index.php/customers/embed/video?' + urllib.parse.urlencode({
  80             'id': video_data['id'],
  81             'user_id': video_data['user_id'],
  82             'signature': video_data['signature'],
  83             'countries': 'Q0M=',
  84             'filter': 'DENY',
  85         })
  86         info.update({
  87             '_type': 'url_transparent',
  88             'url': embed_url,
  89             'ie_key': 'MangomoloVideo',
  90         })
  91         return info
  92
  93
  94 class AWAANLiveIE(AWAANBaseIE):
  95     IE_NAME = 'awaan:live'
  96     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?live/(?P<id>\d+)'
  97     _TEST = {
  98         'url': 'http://awaan.ae/live/6/dubai-tv',
  99         'info_dict': {
 100             'id': '6',
 101             'ext': 'mp4',
 102             'title': 're:Dubai Al Oula [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 103             'upload_date': '20150107',
 104             'timestamp': 1420588800,
 105             'uploader_id': '71',
 106         },
 107         'params': {
 108             # m3u8 download
 109             'skip_download': True,
 110         },
 111     }
 112
 113     def _real_extract(self, url):
 114         channel_id = self._match_id(url)
 115
 116         channel_data = self._download_json(
 117             f'http://admin.mangomolo.com/analytics/index.php/plus/getchanneldetails?channel_id={channel_id}',
 118             channel_id, headers={'Origin': 'http://awaan.ae'})
 119         info = self._parse_video_data(channel_data, channel_id, True)
 120
 121         embed_url = 'http://admin.mangomolo.com/analytics/index.php/customers/embed/index?' + urllib.parse.urlencode({
 122             'id': base64.b64encode(channel_data['user_id'].encode()).decode(),
 123             'channelid': base64.b64encode(channel_data['id'].encode()).decode(),
 124             'signature': channel_data['signature'],
 125             'countries': 'Q0M=',
 126             'filter': 'DENY',
 127         })
 128         info.update({
 129             '_type': 'url_transparent',
 130             'url': embed_url,
 131             'ie_key': 'MangomoloLive',
 132         })
 133         return info
 134
 135
 136 class AWAANSeasonIE(InfoExtractor):
 137     IE_NAME = 'awaan:season'
 138     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?program/(?:(?P<show_id>\d+)|season/(?P<season_id>\d+))'
 139     _TEST = {
 140         'url': 'http://dcndigital.ae/#/program/205024/%D9%85%D8%AD%D8%A7%D8%B6%D8%B1%D8%A7%D8%AA-%D8%A7%D9%84%D8%B4%D9%8A%D8%AE-%D8%A7%D9%84%D8%B4%D8%B9%D8%B1%D8%A7%D9%88%D9%8A',
 141         'info_dict':
 142         {
 143             'id': '7910',
 144             'title': 'محاضرات الشيخ الشعراوي',
 145         },
 146         'playlist_mincount': 27,
 147     }
 148
 149     def _real_extract(self, url):
 150         url, smuggled_data = unsmuggle_url(url, {})
 151         show_id, season_id = self._match_valid_url(url).groups()
 152
 153         data = {}
 154         if season_id:
 155             data['season'] = season_id
 156             show_id = smuggled_data.get('show_id')
 157             if show_id is None:
 158                 season = self._download_json(
 159                     f'http://admin.mangomolo.com/analytics/index.php/plus/season_info?id={season_id}',
 160                     season_id, headers={'Origin': 'http://awaan.ae'})
 161                 show_id = season['id']
 162         data['show_id'] = show_id
 163         show = self._download_json(
 164             'http://admin.mangomolo.com/analytics/index.php/plus/show',
 165             show_id, data=urlencode_postdata(data), headers={
 166                 'Origin': 'http://awaan.ae',
 167                 'Content-Type': 'application/x-www-form-urlencoded',
 168             })
 169         if not season_id:
 170             season_id = show['default_season']
 171         for season in show['seasons']:
 172             if season['id'] == season_id:
 173                 title = season.get('title_en') or season['title_ar']
 174
 175                 entries = []
 176                 for video in show['videos']:
 177                     video_id = str(video['id'])
 178                     entries.append(self.url_result(
 179                         f'http://awaan.ae/media/{video_id}', 'AWAANVideo', video_id))
 180
 181                 return self.playlist_result(entries, season_id, title)